Arsenal 2-0 Watford: By the numbers

35

Arsenal have five straight league wins, a second consecutive clean sheet and are even on points for fourth place.

The above, all sound great. However, while the results have been good, the actual performances have not been great. This is better than the alternative of good play, with bad results. It is great to bank points during these spells, because even if the wins are a bit dodgy and ugly you don’t have to give them back. Now the task of Unai Emery and his staff is to get the players, playing the way he wants.

By the pictures:

Rob Holding makes a case to be a regular starter

5 – Aerial duels won of 7. Tied for most.

9 – Ball recoveries, led Arsenal

0 – Times dribbled past

7 – Clearances, led Arsenal

0 – Fouls committed

3 – Interceptions, led Arsenal

2 – Offsides provoked, led all players

1 – Block (it was a big one, see below)

1 – Dribble completed

89% – Pass completion percentage, 4th best on Arsenal

With Sokratis missing out on today’s match with an injury it was a great sign to see Holding be able to step in and have a great match. He was a very steady performer on defense for Arsenal on a day that Arsenal didn’t look very good on that front. I think that he is making a case to be a regular starter.

Mesut Özil imposes his influence (finally)

65 – Touches for Özil, the most he has had in a match this season

36 – Final third touches, tied for most in the match.

22 – Final third passes completed, second most in the match (counts only passes that start and end in the final third)

15 – Final third passes completed in the second half, led all players (counts only passes that start and end in the final third)

4 – Passes into the box completed, led Arsenal.

1 – Goal

Getting Özil into matches has been difficult so far this season. He has not looked marginalized on the right hand side of the Emery 4231 system. Today he started on the right hand side and looked lost in the first half.

In the first half, Özil had to drop deep to be able to find any of the ball. Arsenal continued to have a huge left sided biases in their attack leaving Özil isolated on the right.

To start the second half, Emery seemed to make a tactical switch with Lacazette and Aubameyang taking turns switching as the right side forward. This change seemed like it was huge in unlocking Özil to run the Arsenal attack.

To end the match, with Ramsey subbed off Özil was free to take up a more free floating position as the number 10.

If Emery is set on keeping Ramsey in the stating line up, the switch of Özil from the right to the left might allow him to still influence the game even if he isn’t in his preferred role.

Watford waste golden chances

6 – Big Chances for Watford, the most that they have created in a match this season. (A big chance is defined by Opta as: A situation where a player should reasonably be expected to score usually in a one-on-one scenario or from very close range.)

3 – Big Chances on target for Watford, plus one blocked. Big Chances on target are converted into goals 70% of the time.

5 – Big Chances for Watford while the match was still 0-0.

1.8 – Expected Goals of those 5 chances.

Here are the five chances, with the explanations that go into the xG for each.

Troy Deeney, 52nd minute. shot from feet, from the center of the box, following a set play, classified as a big chance. 0.43 xG and 0.84 in the on target model.

This was a great save from Bernd Leno, where he has almost no time to react to a ball that was put towards the corner of the goal.

Andre Gray, 68th minute. Shot from feet, from the left box, following a through ball, classified as a big chance. 0.32 xG and 0.4 in the on target model.

Leno does a really good job to come out and make the target small here. I also really liked that he stayed up and big all the way through to the shot. Another really good save.

Isaac Success, 72nd minute. Shot from feet, from the left box, classified as a big chance.  0.22 xG and 0.0 in the on target model (Blocked).

Rob Holding had a long way to go to be in position to get this block. It showed great anticipation as Shkodran Mustafi tried to play Success offside and failed.

Craig Cathcart 74th minute. Shot from feet, from the six yard box, following a set piece, classified as a big chance.  0.49 xG and 0.7 in the on target model.

I think that this is probably a harder chance than how it is rated, Cathcart is off balance here as he is trying to shoot and can’t make clean contact. These are factors that the expected goals model doesn’t know about. It was still good from Leno to stay big and then to some how hold on to the ball after the initial save.

Isaac Success 75th minute. Shot from feet, from the right box, following a fast attack, classified as a big chance.  0.34 xG and 0.0 in the on target model (missed).

This was one of Holding’s only misstep in the match, he was beaten to the goal side by Success who had tons of space to run into. This shot was the closest Watford came to scoring with the chipped shot hitting off the post over Leno.

@oh_that_crab

Sources: Whoscored, Statszone, my own database

35
Leave a Reply

avatar
newest oldest most voted
Ordnance Dave
Ordnance Dave

Really rode our luck today. Unai has to get on top of things before it runs out. Having Ramsey or Ozil, not both, would be a start.

412gooner
412gooner

I think we need to go with Ozil and leave Ramsey out considering the contract situation. Plus moving Ozil central gets the most out of him and bringing in someone like Iwobi/Mhki provides better balance to our front 4.

DamoDinkum
DamoDinkum

412gooner – Yes, yes, a thousand times yes.

Lesster
Lesster

I would agree at first thought but you also have to realise that Ramsey gives us defensive balance as 5 of Watford’s 6 big chances happened when he was off the pitch
Tricky choice

Ordnance Dave
Ordnance Dave

If Unai insists on fullbacks playing like wingers, then maybe we should go with a narrow midfield and two up top. God help us in defence tho.

Richard
Richard

We may have been lucky, but let’s not forget that we should’ve been 2-0 up by the tenth minute (penalty claim + Lacazette missing that sitter)

Tony Adams nose
Tony Adams nose

They can go on and on about how unlucky Watfart were but the truth is they ran out of steam. After we scored they visibly shrunk into their shorts. Then we scored again.

This is not Emery’s finished article by a long way obviously but this game was a big step in the right direction.

Oleg luzhny
Oleg luzhny

No cojones

John O
John O

I wonder if one reason we have been outperforming our stats is that Aubameyang and Lacazette are cli Inca, and finish at a better ate than Opta would expect based on where they are shooting from.

John O
John O

*clinical

1231

I think our calmness and ability to fight late into the match with confidence is something that the stats miss. The quality of our front 4/5 is super high.

That said, today it was an OG and a close range finish so i’m not sure it applies to this game.

Wolfgang
Wolfgang

The defence is still a problem. Against Pool/Chelsea/MC the gunners would be ripped apart by these teams fast moving attack.The gunners attack as usual is too predictable.Incessant passing and probing.
The result is like Wengers time the gunners are taking on a fortress.
Cue. When the gunner attack is stopped,a through ball will stop it and with a goal down the team will throw everything into the attack. Shades of Wenger.Soon it will be 2 goals to the enemy.
Why not vary the attack. A through ball grounder .That will by pass defenders guarding the approaches.
Anyway Im happy for the 3 points.Sooner or later the winning run has to stop.
Why not before the Pool game. Then a win will be most welcome.
COYG.

Bob
Bob

Lacazette’s chances alone how do you work out under 1 “expected goal” for Arsenal…

Mondo
Mondo

I’m not exactly sure how xGs are worked out, but if you’re referring to the on target SoTxG at 0.75, Laca’s attempt was off target so wouldn’t register there.
Each attempt gets a fraction shown as a decimal, and although I don’t know how they would rate the own goal, I think Ozil’s chance would rate pretty high. But maybe Scott could enlighten us on the exact numbers.

Dave
Dave

Yeah I’m becoming less and less inclined to believe this xG model is of any real use. Seems like over half the time it suggests the score should be completely different to what it was (3-1 Watford is a massive turnaround)

Lacazette was clean through on goal for a bigger chance than any of the above Watford ones were, yet it doesn’t seem to register as high.

I look at the above examples and don’t see how the Andre Gray one or the Cathcart one can be classed as big chances. The Cathcart one was even disputed in the article. Just because Gray eventually got to the ball in that position doesn’t mean it was a clear chance. xG doesn’t seem to account for a player stretching or the pressure he is under etc.

In my most cynical view I start to wonder if xG if fundamentally flawed to allow “opinion” (read “bias”) into it’s calcs. Obviously no model is perfect though and I’m obviously pro the Arse so maybe I’m just being a grumpy man.

Scott Willis
Scott Willis

Single shot xG should have large error bars around it because the inputs are all generalized from similar chance but are not the same. Over the long term they balance out but certain chances can easily be over or underrated pretty easily.

Scott Willis
Scott Willis

You can think of the decimal representation as a percentage that it is scored.

The first chance was 0.25, I think it was better than that but it was outside the box and distance to goal is one of the biggest factors. From the on ball data the model doesn’t know it is one on one.

Bob
Bob

Doesn’t that point to the limitations of analysing performance by looking at xG?

tw_ht
tw_ht

Of course it does. But just because there are limitations to xG doesn’t mean that there is no value to the metric. Football is a low scoring sport with a large amount of luck involved. You need “chance quality” (i.e. xG) metrics in order to sort out which teams are riding their luck versus actually performing at truly high level. Like we saw today v. Watford, the final result doesn’t always match the chance quality within the match but that’s ok. What’s the alternative? Watching every match in every league and subjectively rating every single chance that appears? Impossible. We need models in order to consistently compare teams across entire seasons, between leagues, for long stretches of time – now and into the future.

Furthermore, there are groups (like StatsBomb) that are working in order to eliminate some of the problems xG has with chances like Lacazette’s. Their model takes into account the position of every player on the pitch (goalkeeper included) for every scoring opportunity. So in a situation like Lacazette’s where he was 1v1 with the goalie, their model likely would grade the chance as higher than 0.25, even if it was relatively far away from goal.

Bob
Bob

I think there’s probably an old fashioned solution where we just judge the games based more on what we saw rather than relying so dogmatically on xG and other statistics maybe?

ach
ach

Feels like Ozil actually went missing when he moved centrally until he scored the goal. Emery system for some reason doesn’t seems to involve the number 10 much. Ramsey went invinsible all season, ozil made only 20 something pass all games againts chelsea when he started there. Maybe it’s because the lack of natural winger making the center congested.

PEA14
PEA14

Iwobi have seems to improve alot, showcasing his ability with the ball under his feet. He has also shown improvement in his defensive duties and a quicker mind too. He should be given a run of few matches for his efforts

Nameless gunner
Nameless gunner

How abt adding some amusing stats? This section is very technical these days..

Goonarf
Goonarf

Just one mans opinion here. The glut of statistical references weigh down these articles for me. The math, numbers, decimals etc get in the way of the basic numbers telling a story. I much prefer the style of presenting the basic and digestible stats and then opining on whether they match the narrative or not.
This article used to be fun to read. Now it feels like homework.
All that said, I sift through for the bits I enjoy and do appreciate the effort it takes to produce the article. Just my two cents.

GraeB
GraeB

Totally agree. I made the comment a few weeks back but have stuck with it to see if it gets easier or helps me understand the game any better – sorry, it doesn’t.

tw_ht
tw_ht

To be blunt, try harder then. Just because something doesn’t immediately make sense to you doesn’t mean it doesn’t have value in understanding the match and sport at large.

Gooners & Roses
Gooners & Roses

I have to agree, By the numbers from Tim 7amkickoff was one of my favourite section before. It was fun for the layman, and highly insightful. I try to enjoy this new version but maybe the new style doesn’t fit me.

SeryxMe

I’d say it again, this idea of letting the opposition so many opportunities to score without scoring is becoming a(n unwelcome) skill. Like the the Gent said on Friday, it’s like we offer the visitors goal-scoring ‘refreshments’ before we eventually put them to the sword. I absolutely love that we’re winning without being at our best, however, the key concern is that we can’t be this lucky against the top teams as shown by our first two games.
In any case, our goalkeepers seem to be in top form and our frontline is getting us the goals, so let’s just hopefully keep getting better while we’re garnering the points before our next big game. We’re still on course for our 24pt haul! COYG!

Garmouth
Garmouth

Picking up on what others have said, it is becoming clear that there are fundamental flaws in xG models because they do not, as they are generally understood to do, measure the goals you should expect a team to score; they only measure the quality of chances created. All the top teams in the Premier League can put out £100m+ of goalscoring talent and all have goalkeepers who are in the 10 most expensive of all time (or would be if they were bought now.)
So when a team like Arsenal comes up against a team like Watford whose forward line is led by a man would come second in a “hitting a cow’s arse with a banjo” competition, the weakness in the model becomes clear. And that weakness is exacerbated by other systemic issues like not considering who the chance falls to within the team; surely a defender is less likely to score a given chance than a striker. Similarly, own goals seem to be discounted when clearly something like Arsenal’s first is a result of good attacking play both in terms of getting a wicked cross in at the near post and Laca’s aggressive play panicking the defender. I’d say (totally unscientifically) that there must have been at least a 50% chance of somebody turning that one in, deliberately or accidentally, and yet where is it in the model?
Statistical models can be very useful, especially when they help us to reconsider our preconceptions; the move to OBP as a key metric in baseball is a good example. But when the model delivers results that are clearly out of line with perceived reality then, at the very least, the model needs to be questioned. And a model that says a 4-0 Watford win was more than twice as likely as the actual 2-0 Arsenal win is, in my view, a long way across the border into that questionable territory.
This is not to say, though, that the model is worthless – clearly giving Watford so many good chances is pretty poor and indicates that you will probably be in trouble against better teams – but that it is presented misleadingly. Call it ‘quality chance creation’ and it will do what it says on the tin.

Olivije Žirod
Olivije Žirod

I am glad that we have stopped playing out from the back for all cost and we are picking more direct approach. I don’t know if anyone noticed that but since Newcastle game we stopped forcing that playing style. Against Everton if I am correct Cech had half less passes then in previous games, against Vorskla and Brentford I don’t remember to much playing at the back and the same for yesterday. Even when Leno came on we didn’t do it. Maybe because of Holding?

Going forward we play more possession football (which is more similiar to Wengerball) than in the first matches. Maybe it is a change of tactics or just Emery’s ideas gluing together.

Bob
Bob

I’m not sure we have, I think it seems to change depending on the sides we play

Emperor
Emperor

I keep hearing people criticizing the xG, this is something you must understand, statistics do not and cannot tell the full story, they only tell you a story. What’s the story? Taking into account the number of goals scored and missed from a certain position on the pitch, it calculates the probability of scoring a similar chance from the same position. That IS the story it’s telling, it’s not supposed to tell you the whole story

Garmouth
Garmouth

Of course, statistics don’t tell the whole story but I think you are missing the point. The xG is based solely on where, relative to the goal, the chance is created and then applies to that how likely it is that the average player, facing the average keeper will score. So it applies the same probability whether the ball goes to Lacazette or Holding, and whether he’s trying to score past David De Gea or Lee Grant. I’m not saying it’s a bad statistic, it isn’t. But it is a very misleading one if you think it is a realistic measure of expected goals which is how it is presented and, I think, how it is generally perceived.

mike
mike

I think something is wrong with the Ozil touches charts. All three are only showing one red dot. I assume there should be 63 dots across all of them.

Xxxrob
Xxxrob

Just want to say, no disrespect to mr Stillman, this segment hasn’t missed a beat since his departure. Great stuff.